My improvised planetary camera

[megalomania]

As I may have mentioned earlier I am building an improvised planetary camera along similar lines as of the Bookeye planetary camera. For those who don’t know what a planetary camera is, it is essentially an overhead digital camera that takes pictures of a book in the books normal reading position. Planetary cameras start at $20,000 and go up from there. A digital camera can do basically the same thing with a bit of work and minus the fancy software.

I have just completed scanning my first book with this method, the 63 page “Explosive Principles” by Robert Sickler which is uploading to the FTP as I write this. This is actually the first book I have scanned from start to finish :) Its small size seemed perfect for my tests.

Scanning the book was a snap. All told I scanned all the pages in about 3 or 4 minutes. Setup required less than 10 minutes, but considering this was my first time I expect this to take less. I used a 1 foot threaded rod attached to the cameras tripod hole and a lab ringstand to hold the camera. I leveled the camera and focused in on the book. Once that was done all I had to do was snap the pic and turn the page.

Post processing took a few hours, but again had to download software, learn it, and use it for my first time. Now that I have a handle on things I could probably do everything quite quickly. I used batch image processing software to rotate/straighten some images, crop the margins, convert to grayscale, crop again to split the two page images into one page images, and rename all the files to even and odd to correspond to the page numbers. Finally all the images were converted to PDF files and saved into one single document.

I made many mistakes and discoveries along the way that I am writing a web page about. When I am done I will make a detailed step by step how to guide complete with pictures for others to follow along with me. I want to know if anyone out there has had a similar experience to mine and wants to share their insights while I write the page.

The seeds for this project were planted several years ago while I was in the library. Being a cheap bastard I refuse to spend $0.20 a pop on photocopies, and I don’t particularly care for writing down the contents of sozens of pages of material. Some books can’t be checked out so what to do. If only I had a digital camera I thought… Now I have a nice 4.1 megapixel camera with a 128 MB card. I still didn’t make the connection until I accidentally stumbled on a webpage about the bookeye. Check out <a href="http://www.digitise-it.com/scanners/scan2.html" target="_blank">http://www.digitise-it.com/scanners/scan2.html</a> to see exactly what I am talking about.

I estimate the time to scan one page of a book to be 10 seconds each. This includes turning the page, taking the picture, and downloading the images from the cameras memory card. If a picture of two pages at a time is taken this time can be halved. At 5 seconds a page one could scan 1440 pages in 2 hours! (a large book in about the time I can reasonably be expected to sit in one spot nonstop) Post processing of the images will still be required in addition to this, but you have to do that no matter what method you use to scan. A real planetary camera can take the picture, split the pages, assemble them into a PDF file and run OCR all in real time even while you are taking the pictures. I would expect no less for $20,000 though.

It will still be a few weeks before I finish writing my paper about this. In the mean time I hope people would be wiling to share how they process their scanned images? Up until now I have only scanned individual pages straight into pdf files while being anally sure the page was aligned while it was being scanned. This takes a lot of time this way. Also I did not compress my images or anything. The individual images of my latest book were about 125k a pop after cropping and reducing to grayscale, and the book is 7.81 MB for 62 pages. I don’t know if that is too big or what. I still have to apply OCR to the file, the two programs I have are not yet installed. I also forget how to apply compression on pdf files, it has been a long time since I last did that.

[nbk2000]

I wonder what the minimum camera resolution is to make this work? I could buy a 2megapixel camera at work for about $130, and hook it up to my computer to use it to capture pictures directly into it, without using the card.

[megalomania]

I use 1600x1200 resolution at high compression on my 4.1 megapixel camera and I get about 280 images per 60 MB. This resolution is about half, or two thirds of what the camera is capeable of, so I estimate a 2 megapixel camera may just do the job. The quality of images I get seem to be on par with 600-800 dpi scans, and that is 2 pages at a time.

I actually reduce my image quality by 60% of the original and I still get a clear image more on par with a 300 dpi scan. I did find that setting my camera on 640x480 resolution was unacceptably low.

Of course I have no other frames of reference to gauge if a lower megapixeled camera will work or not. What information that is on the Internet is very sketchy. Of course considering I get good images even after reducing my quality by 60% and that from the middle setting I don't think it is too far a stretch to assume a 2 megapixel set to maximum quality can do the same thing.

I am actually considering buying one myself around Christmas (a 2 MP camera). The owners of the camera I use are less than enthused at having me taking endless thousands of pictures, and then they always seem to be using it when I want it (owning a digital camera turns everyone into a shutterbug, you come to want it handy all the time to take pictures of everything).

You have to watch what camera you get though because not all of them can be connected to the computer in the way you want to use it, nbk. Only more recent cameras have the feature, and likely the more expensive ones. Mine for example can only transfer images through a USB cable, but it can not do so at the same time an image is being taken. The camera has to be physicially configured to set it to computer mode and a special spftware package run to download the images. I actually pop out my card and use a card reader that is part of my printer to do this job far more efficiently. This still renders the camera ineffective during the lengthy process.

I am sure you meant have the image transferred directly to the computer while you take it, rather like a webcam. Again, that is something you have to look for before you buy because not all cameras have this feature.

[Rhadon]

Mega, it's quite a long time ago since you wrote your first post, but it's only now that I see you're asking for others to share their experience. I had the idea to create a PDF that teaches others how to create good-looking PDFs with minimum effort using a flatbed scanner. But this is a project for the future, I've got other things to do right now.

On http://www.digitise-it.com/scanners/scan2.html they mention a program which can do a "fold correction". That one caught my attention and would be a very nice thing to have. Perhaps someone has an application which can do this or can at least name one. Thanks in advance.

[megalomania]

I have looked into the program you mentioned, I believe it comes with most planetary cameras. Unfortunatly this software cannot apply such fold corrections after the fact, it has to be corrected before the picture is even taken. Since digital cameras, or even scanners, arn't equiped to deal with this type of technology, the software won't work. It has to be coupled with special hardware that is part of the imaging system.

It would seem to me that a simple photoshop plugin could do the job, but apparently there is some hugh technological hurdle that prevents images from being flattened and uncurled in this way. A pity.

Nevertheless I tried to find the software, but it would be very rare as it is a $700 item, and usually only bundled with a $25000 unit.

[Rhadon]

Flattening the image may seem easy at first, but in fact it's a rather complicated job because the depth of the fold differs from book to book and even from page to page. For that reason the plugin would have to analyze the page and find out about this details on its own which is a hard thing to code. One would be most likely to find such functionality in OCR programs, but since the new FineReader 7.0 doesn't have it I assume that we'll have to live with the fold. Thanks for taking your time!

[IDTB]

This is alittle off topic, but it might be good information. You decide.

I believe it would be alot easier if you simply use a program that analyzed the pictures rewrote the letters it sees in another document. It would be silly to have a huge file due having data presenting the pages/book when you could simply have data only presenting the print.

I'm sure finding such software or even creating it shouldn't be that hard. It would work mostly on comparing what it sees to it's database of characters.

I'm not sure if I make any sense. If I don't, tell me and I'll try to elaborate as much as possible.

[nbk2000]

The camera I'm looking at has a 1,600x1,200 pixel image (actual). It will hold 25 images on 16mb of storage card, so how would I figure out DPI from that info?

[IDTB]

I believe you derive the DPI from the megapixels.

[nbk2000]

Boy, that was informative. :rolleyes:

[Anthony]

IDTB, you might want to look up "OCR":p

EDIT: Surely if you took an image 8"x6" at 1600x1200 that would be 1 920 000 pixels, spread over an area of 8x6= 48sqin = 1 920 000 / 48 = 40 000dpi

Seems very wrong...

[IDTB]

I was informing you how to figure out the DPI, because I didn't think you knew. You said, 'from that info?'. I don't believe there's anyway to figure out the DPI from that, as I informed you. If you provide adequate information, I too will provide an adequate answer.

I'm sorry if you didn't find me helpful.


Thanks, Anthony! That's exactly what I was trying to think of. ;)

[a_bab]

Anthony, your calculations are *almost* exact. The problem is that dpi stands for Dot Per Inch, and not Dot Per square Inch, as in your calculation. So basically what you've got there is the number of pixels you have spreaded over a square inch. You only need the number of a frame of your square, so since it's a square you can extract the root out of your area number, which is 40,000.

200*200=40,000

In conclusion, 200 dpi.

[A-BOMB]

That is some cool stuff Mega I think I just found a use for this 1.3Megapixel aiptek digicam that I have I think I should be good for semi-good resolution. I have this spy book that describe this process in makeing microdots and photo copies I'll see if I can find it and get some pics of it.

[Anthony]

Thanks for the correction a_bab, I see it is obviously dots-per-inch on a line now :)

[nbk2000]

This is how I figured it out.

Image size in pixels

H x W = Pixels

1600 x 1200= 1920000

Pixels / Image byte rate (24 bit RGB =3 bytes per pixel) = Image Size (bytes)

1920000 / 3 = 640000 bytes per image

This agrees with the specs on the box for both image pixel size and for image storage capacity for the included card. (16mb/640kb=25 images).

Image size (bytes) / Image size (sq. inches) = X

The square root of X is the DPI for that image.

If we figure an 8" x 5" page image, that's 40 sq. inches per image.

640000 / 40 (square inches) = ~120DPI

A 3" x 5" image would be:

640000 / 15 = ~205DPI

Assuming a minimum of 150DPI for passable OCR, that means a book page no larger than 28 square inches in size could be processed by this camera. :( So I'll not be buying it.

If it was possible to take the pictures in B&W, than that'd make it feasible, because 8 bit Grayscale uses only 1 byte per pixel, and 16 bit Grayscale uses 2 bytes per pixel. However, the camera doesn't take pictures in anything BUT color. :mad:

[nbk2000]

Well I bought the camera anyways, since I can always return it.

I took a picture of a typewritten page and cleaned it up in photoshop, adjust the resolution, converted to B&W, saved to TIFF, and opened it in my OCR program. It worked. :cool:

Now to try something more complicated like a magazine page. There may be hope yet. :)

++++++++

Good news, bad news.

Good news is that it'll do for a small book that I'm working on that has several hundred pages of single column text. :)

Bad newsit can't do magazine articles for shit 'cause the type is too small and over too large an area. :(

Though, if I just take a picture of the page, it's readable to the eye, even if OCR can't make heads or tails of it, but with a small bit of fish-eye distortion at close range.

[Jacks Complete]

I have tried to do this, using a 4 megapixel camera, and had very little success.

I was trying to do "The Crossbow" by Ralph Payne-Galwey, cos the seige engines are brilliant, as well as the info on different systems and all the history of the crossbow.

Anyway, what OCR packages have you guys been trying? I got nowhere with the free one I got with my scanner, even after I hacked it to take files as well as scans. It was Textbridge Classic 2.0, but it just didn't work, as it couldn't cope without lots of photoshopping first.

What would you recommend?

[peterthesmart]

In the latest issue of popular science, they had an picture and description of a planetary camera like device for about $50,000 that could scan 1,500 pages in an hour automatically. I'll scan the article when I find some time. Better start savin up!

[Rhadon]

Anyway, what OCR packages have you guys been trying?
In this (http://www.roguesci.org/theforum/showthread.php?threadid=1672&postid=48007#post48007) thread I wrote how I scan my books. You should get FineReader if you can, it's much better than Textbridge.

[megalomania]

I am getting ready to improve my process a good bit and take a cue from the real planetary camera people. My two main problems include flash washout and how to hold sown the pages of some books.

I think I have both problems licked, we shall find out. The flash washout, I believe, is caused by my flash being too close to the page. It is far too bright in one spot while it gets progressively darker towards the edges. This has also caused some trouble when I convert the images to black and white as the lack of contrast between the black words and the white page where the flash was brightest causes those words to be unreadable. If I pull my camera back about 12 inches it should diffuse the flash enough to limit the washout effect.

This creates two problems though. First I will need a stand height approaching 3 feet tall, and second it will mean the edges of the page will be darker still. Sometimes even with the flash close the corners can be unacceptably dark and unreadable. I plan to use some halogen floodlights on either side as bias lighting just like they do with planetary cameras. Hopefully this will be bright enough to lighten the edges of the page whereas the flash will brighten the center. I may even go so far as to build a cardboard shell and cover the inside with aluminum foil to reflect as much light as possible on to the book pages.

The stand height is easily taken care of. I have a large capacity ringstand without a rod, I will have to buy a rod at the hardware store and just glue it in, or thread it myself. An alternative I have been toying with is to use some gooseneck lamp arm, or one of those mechanical levered lamp arms that let you swing the lamp into any position. Using a gooseneck might simplify positioning the camera rapidly because it can make fine adjustments quickly. Using a ring stand is rather hit or miss requiring lengthy readjustment. I do worry that neither a gooseneck or mechanical arm will not be able to support the weight of a camera, nor will it stay in place with the constant movement of pressing the camera button. I will think of something...

My next problem is how to hold the pages down. Using a pane of glass worked great for this, but the reflection was horrible. I can't not use the flash, the resulting text is unreadable. I have extensively researched information on anti-reflective glass. I have concluded it may be too expensive. From what I can gather the best source of anti-reflective glass is framing shops. The sell crystal clear glass that looks invisible for framing artwork. This glass is coated on both sides with materials that reduce reflection down to around 1%. Normal glass reflects 9% of light. There are no coatings, sprays, or films that we as consumers can apply ourselves, there are no cheap alternatives.

Anti-reflective glass has a few brands that are used, Denglas, TruVue, Image Perfect, and Luxar being the major brands I could find. None of the manufactures sell by the piece, and very few framing shops online sell the stuff mailorder. Fortunately there are many framing shops in my area; someone should be able to get me some assuming the price is not too high. It may cost as low as $85 for a single piece of 18" x 18" glass! Acrylic plastics cost even more, but I need the glass for weight, so plastic is not so good. I may not need to go as high as 18x18 of course. 18 inches wide would be enough to cover both pages of an A4 sized book (8.5 x 11, like a sheet of paper), but come to think of 18 the page curl is way too excessive on such large books. Most books that size tend to be thick, or at least the ones I will be scanning, and that means I will have to take one page at a time. Something closer to 12x10 would be best for smaller, thinner books. I didn’t quite catch what the lady on the phone said, but I think they charge by the square inch.

[A-BOMB]

Mega you can try this, turn off the flash on the camera and use 2 bright lights that are off too the sides of the book you are taking a picture of. I did this when I tried this with my camera I took my tripod and attached a piece of wood to it then I drilled a bolt into the end that was threaded to fit the attachment on the bottom of my camera, I then put a counter weight on the onther end of the wood rod. Then I used the 2 desk lamps I had next to my computers I aimed the lamps and adjested the focus on the camera and it turned out pretty good. The 2 lights evened the light out across the pages. Just make sure that your useing good lightbulbs that have a uniform white light in each lamp.

[inFinie]

If anyone have matlab:
In this link http://www.math.rutgers.edu/~ojanen/undistort/index.html
it covers two types of distortion,zip file contains an exe file disappointingly it needs matlab to compute some parameters.
EDIT: A plugin might work: http://www.pluginsworld.com/Photoshop/plugins.php3?action=software374&soft=Photoshop

[megalomania]

Yes, A-bomb, that was what I mentioned in my last post I was going to do. I set up my system today and it worked quite well. Those 2 halogens throw off some heat, good thing it's winter so I can open a window. I used the glass window I have to hold the pages down and I got no reflection. It took a little fiddeling with my camera exposure settings to get it just right. Normal exposure is no good without a flash so I set it at +2 EV. I am not quite sure what that means, but I think it keeps the shutter open a little longer. All the other exposure settings left the image black.

The final picture was still a little too bright for ClearImage to render into legible text unfortunatly. I would move the lights back a bit but I am out of table space. I have a countertop in the garage, I'll have to use that to give me some width.

I have attached my sample image to show my results, I hope it isn't too big...

[megalomania]

I have completed my latest rounds of camera tests and I finally found a system the lets me use ClearImage to render the text readable. I elevated the halogen lights about a foot and angled them so they point more at each other than on the book. This creates a substantially darker image, but when converted to bitonal in ClearImage all the text is readable.

Before only the text on the inside margin, farthest from the light, was readable. I could probably raise them a little higher to reduce the light just a tiny bit more to make the text even better.

I found out that I do not need to write a vbasic script to run a batch process in ClearImage, I only need to make a bat file to run it. I still have no idea how to make a bat file, but it looks as if it may be easier than programming vbasic. If anyone knows anything about making a bat file please tell me. I want it to open each image one after the other, apply deskew and autocrop, and then save them as tiffs.

Attached you will find a zip that has the original jpg I took, and then the processed tif file. The tif file was reduced 86% from the original. I converted it to a pdf file, and then ran jbig compression. Unfortunately the silx jbig compression actually added 3k to the pdf. Maybe it will not do that if I have hundreds of pages to compress.

[Anthony]

A flash-diffuser might have helped ypu to avoid the washout in the centre of the page. Something like a piece of tissue paper taped over the flash. But considering that the edges were already under-iluminated, it probably wouldn't have helped.

[Bert]

Have you considered building a light box? Make a square box of transparent plastic or paper difusors, put the lights outside of it on all sides shining in and the object to be photographed inside- This provides difused light evenly from all sides. Turn off the flash and go with the now bright, apparently sourceless ambient light inside the box. Should allow use of an ordinary glass plate to hold down pages as there is now no point source to cause reflection.

[megalomania]

Using my twin halogens I can now get exactly the right amount of light, after adjusting the timing on capture speed. SInce the camera was not getting anough light I increased the length of time the picture is taken and the amount of light is drasticially increased. At +2 the image is quite bright, at -2, even at +0.5 the image is pure black. By varying the direction the lights are pointing (directly on the page vs towards each other) and by varying the distance the lights are to the book I can get different light levels.

I have two light settings now, one to get a good quality color image that is human readable (full light, and +2 shutter delay) and the other to get a machine readable image able to be converted to bitonal and still be machine readable (lights angled towards one another somewhat farther away, and shutter lag set to +1 to make a much darker image).

The human readable images are best for color pages with graphics and colored text, or any old graphics for that matter. The machine readable text only becomes human readable after processing, its advantage is a significiantly reduced filesize. Graphics are, however, less than readable.

[Bert]

I applied for FTP access about 2 weeks ago, hope to get it someday soon. I've got a digital camera (Olympus Camedia C-3000) with a tripod mount- I'd be interested in playing with this technology and these programs. I've got a few books you all might be interested in if I can get the techniques down. More to learn about the computer...

[nbk2000]

Nothing beats direct sunlight for brightness and diffused source. Though this might make planetary scanning a seasonal activity. ;)

[chemwarrior]

Check your mail again Bert. I sent you the login and password a while back...

[Mike76251]

If the light reflection problem persists off your glass you might look into a polarizing filter over your lens. One with two pieces of polarizing glass that you can rotate will knock at any and all reflections off glass.

[megalomania]

Now that I use flanking lights I don't have a reflecting problem, although a polarizing lens is a good idea I never thought of. That would certainly be cheaper than a pane of anti-reflective glass.

[megalomania]

I have had a breakthrough recently in my technique. Lower resolution images, or compressed jpgs incorrectly render black text as varying shades of dark gray. While this is unnoticeable to the human eye, subsequent conversion to bitonal images changes those gray pixels to white. Bitonal images keep only black pixels, everything else gets deleted.

By using photoshops color replacement, actually by varying the lightness of the image, after conversion to grayscale I can darken all the pixels. They dark grays are converted to black wheras the light grays are converted to dark gray. The image is not entirely human readable, but it is not intended to be. Subsequent bitonal conversion resuts in strikingly readable text. Previously I was only able to get such text using special light conditions and no compression of the image. No compression means I can take fewer pictures before I have to download the card to my computer, and this is not something I wish to do. Compressing the images means I can take 4 times as many pictures before the card fills up.

Due to the way my light source plays on the pages it creates a circular pattern that gets progressively darker towards the edges. In grayscale this looks like an increasing circle of gray, but is not very noticeable. Unfortunatly in the very corners the gray is the same color as the pixels I want to turn black. I could compensate for this by adding mirrors to reflect extra light in each of the corners, but that adds an extra bit of complexitty I would like to avoid. The end result are black cresents in each corner that obscures some of the text.

Instead I use photoshpos marquee selection to highlight only a portion of the page. I select a large oval that excludes the corners. I then apply the darkening to the central portion of the page. The corners are already dark enough in most cases to corretly convert to bitonal and be readable.

Once darkening is done and the image is converted to bitonal the text is clearly readable. Without darkening the bitonal conversion makes almost illegable text. I have not tried OCR as of yet, but I expect it to have a high degree of accuracy. I was able to make a macro on Photoshop to do all of this automaticially.

There is some speckeling (noise) that the darkening brings out that I can easily remove with some other software. The undarkened corners do appear a little hard to read sometimes after conversion to bitonal. I may readjust my macro to apply a less intense darkening on each corner. Now I just need to get the software that applies the bitonal conversion to do a batch process. I was thinking of using a macro program to open each image file one by one, apply the processing I want, and save the files. The only problem I forsee there is the image software I want to run the macro on saves every file as *.tiff by default. You have to type in what you want * to be and I don't know if a macro program will be smart enough to add a sequentially increasing number.

[ardjek]

While searching google about copying articles and pictures with digital camera, I came across some usefull information about how to straighten the text of documents and a few portable stands that could be used for copying materials say from a reference library.

Searching "+digital camera +genealogy" I came across
http://www.rideau-info.com/genealogy/digital/copying.html which talks using a digital camera and a reversed tripod to copy pictures.
http://www.csigizmos.com/products/photography/photostand.html has information about a portable PVC stand. It may be light, but steady? not sure.
http://www.magiscope.com/stand.htm
Wood camera stand.

http://genforum.genealogy.com/tips/messages/2459.html it is a collection of articles about using a digital camera to copy documents. The articles are in .txt format. Skim the texts until you find the good stuff, usually around the middle of the text file.
Check out Vol1. No11 to Vol1. No.13.
"Vol. 1, No. 12 - 21 November 2002: The Digital Darkroom, Part 2- a document editing exercise, straighten text and correct density using mask and transform tools, etc."
"Vol. 1, No. 13 -05 December 2002: The Digital Darkroom, Part 3 - editing photos captured from a microfilm reader display, exercise
dealing with common problems including hot spots and image distortion."
Here is a quick link to the text correction article. (The article in the .txt file is more detailed and teaches step by step.)http://www.NGSgenealogy.org/upfront/112102/
Vol 1. No. 7 and Vol1. No 9 have information about making portable copy stands and flash diffusion.
Copy stands linkhttp://www.ngsgenealogy.org/upfront/091202/pubs/

Hope those links help.

[megalomania]

It has recently come to my attention that there is such a thing as panoramic stitching software that will combine multiple images into one. This could have valuable implications for the planetary camera technique because it may be able to jump that last hurdle of allowing high accuracy OCR. Currently a 6 to 8 megapixel camera still gives about 200-250 dpi scans. Stitching two pages would just about double that.

I tried the technique on a new piece of software that is making waves in the photographers sect called autostitch. It just came out last month, is something of a beta program, and lacks documentation for now. My tests with it have shown autostitch to be quite good, better than the few other major stiching progs I tried.

Unfortunatly it left a slight misalignment on my test photos. I suspect this could be eliminated by adjusting my photographing technique. The software also lacks batch capeabilities, but that can be compensated for by using macro software. The lack of documentation is rather disturbing as this software has a number of advanced features that are a complete mystery. At least it is free (for now).

For those interested in giving this a whirl go to http://www.autostitch.net/

I was thinking about using a heavy board to tape my book to, and then tape the board to the table. This way when I finish the top of the page all I have to do is move the board rather than uproot the entire book and risk rotating or distorting it.

[megalomania]

While updating my planetary camera how-to guide I came accross a very recent development in scanning technology. Just last month two scientists at Xerox have published a paper at a conference about a mathematical formula that compensates for page curl and light variations in the gutter of a books spine. That means no curled text or black bar running down the middle of the book.

I don't know if this new formula would work with digital cameras, but there is hope that the problem is being worked on. To read about the new technique see http://www.xeroxagents.com/pages/detpages/xeroxnews492.shtml or http://www-cs.ccny.cuny.edu/~wolberg/pub/xeroxStory.html

[Kamisama]

Now, I like this whole guide and idea stuff. It's kinda funny how you tripped over the idea mega, because I've pondered the idea of how to make quick book scans.

ONTO THE QUESTION

I've been saving up money for a 4.0 MP camera ever since I read the how-to guide. The thing is, I came looking for this type of thread and noticed that you had said it would be somewhat O.K. for a person to use a 2.1+ MP camera. I really question what the megapixel resolution required is.

I've got some books I want to scan 8/9.5 x 11
These books aren't highly graphic except this one (http://www.amazon.com/gp/product/007298936X/qid=1133722665/sr=8-1/ref=pd_bbs_1/104-4060590-4008740?n=507846&s=books&v=glance). It includes photographs and art of different people from Picasso to a detailed B&W picture of Charlie Chaplin in a scene from Modern Time where he is standing next to a bunch of gears (http://www.stadtkinowien.at/imgs/filme/26/01_big.jpg).

In a science book I would assume those gears and their location would be of importance. I do have some psychology books that talk about neuroscience so I find capturing some detail to be important. I don't need great in depth, but I need enough detail to understand what the picture is telling me.

Math books will be of importance; Electronic books, also.

So I've got about $130 USD. I could have bought a 5 MP camera for $45 with the LCD broken but didn't because I did not know if the LCD would be important for the accuracy and position of imaging. I didn't know if some users here found the LCD to be a huge help or not.

If I can get a digicam with a lower MP and if it works correctly, then I should be able to save some cash. However, does a lower MP work?

Note: I don't really care about OCR'ing the text. I don't care about having a search feature at all really. I just care about bending my physical books up because I have to flip through them. I have a page down key and quick eyesight, I'll do fine without the search option.

I figure since I don't care about OCR, that means I'm allowed to have the MP go down a bit, right? I remember you had said in the tutorial that 4.1MP was the minimum for OCR'ing, but I don't feel the need for it.

So from the info, would getting a lower MP be allowed so I can save some cash? Should I go with 4 MP even though? I thought about going up to 5 MP, but as I said I don't care about OCR, I care about image quality and detail.

Quality images like this would become necessary to view:

http://www.ncbi.nlm.nih.gov/books/bookres.fcgi/mboc4/ch1f19.jpg
http://www.ncbi.nlm.nih.gov/books/bookres.fcgi/mboc4/ch1f18.jpg
http://www.telemedicine.org/BioWar/bw06a.jpg
http://www.telemedicine.org/BioWar/bw12a.JPG
http://www.telemedicine.org/BioWar/bw43ba.JPG

Also something more about the turning of pages.
In my head I concieved the idea that a person could build a hinge system with a glass mirror attached, somewhat like on an entertainment center cabinent. That way the could lift the glass, turn the page, lower the glass. Anyone try this?

[justme]

I have been UTFSE for an answer to the question posed by Megalomania in another thread regarding the format for batch or .com files to automate the functions of Clearimage.

Was there ever any resolution to this question?

I have tried to discern how this is done from the help files and the programmer's web site, but unfortunately, I am as close to illiterate in programming matters as it is possible to be.

I am attempting to take the two page scans of "Preparatory Manual of Black Powder and Pyrotechnics" and make it ready for OCR by dividing the pages and cleaning it up with Clearimage.

None of the steps are difficult, but there are numerous steps, and the number of pages involved makes the task seem a bit daunting.

[megalomania]

I ended up contacting the Clearimage tech support, but they referred me to a sales rep who basically would only discuss technical details if I purchased the software. I ended up using a macro program to follow my mouse clicks and to switch between files. I made a text file with all the file names I wanted the files to be named, used the macro to copy a name (I forget now, but through a combination of hot keys including home and something to select the line), switch over to Clearimage, paste the filename when saving that image, and opening the next image on the list. This works 99% of the time. It is only when windows screws up when something pops up, a window switches that should not, or Clearimage crashes that things go wrong.

[megalomania]

Just a few months ago the good people at Atiz released the BookDrive DIY, a do it yourself book scanner that relies on digital cameras. Yes, well, we have been doing this all along for the last few years. What is nice is the relatively low price of $4500, and you supply your own cameras (Canon digital SLRs are only supported so far, like the Rebel XT).

The price is still a bit exorbitant, although it seems the aim of Atiz is to produce these proof of concept products with the ultimate goal of producing an affordable home scanner for ordinary people to digitize their library collection.

The BookDrive DIY is mainly an expensive digitally controlled lighting system with a dual camera USB interface and some software. This is, in my estimation, where most of the money goes into this product. All of that I can do without.

The real innovation that has inspired me is the brilliant, yet simplistic, book cradle system. The DIY can be seen in action at the atiz website here: http://atiz.com/bookdrive_diy.php
Specifically, in the attached image, the DIY uses an angled book cradle that slides on a movable base. The clear platen presses down in the center of the book, and the movable cradle always centers the pages so that it remains in exactly the same distance from the camera for every picture!

It took me a few minutes to realize the implications of this simple innovation, so let me explain further why this is important.

First, the angled cradle minimizes page curl by keeping the book as close to 90 degrees as is feasible. A book opened flat has a considerable hump, or curl, at the spine that makes OCRing difficult, if not impossible. This is a problem that plagues scanners in particular because you can’t angle the book and still be able to scan the page.

Second, the clear cover is very handy to keep the pages of the book from slumping inward due to its own weight, which also distorts the pages.

Third, and this is the most important feature, every time the platen is lowered it forces the movable book cradle to center the book. One of the biggest problems I am facing with my process now is the “page creep” caused by the book slowly getting farther and farther away with each turn of the page. If you have a 1000 page book, page 1 is much closer than page 1000 by 2-3 inches. If you focus your camera to center on page 1, by the time you get to page 1000 the actual area of page your are photographing will be much larger, and this reduces the effective DPI because you are now photographing the margins, the table, and a lot of extra crap beyond the book itself. The only valid solution is to periodically readjust the camera to refocus on the pages, but this wastes time, alters the focusing and lighting characteristics, and plays havoc with post editing.

The post editing is the real bugger. If every page you photograph gets progressively smaller, you can’t effectively do a batch crop and get the margins right every time. As an added bonus, just to make life miserable, photographing a book flat causes margin crawl* from left to right.

* margin crawl is difficult to explain, suffice it to say that the book “walks” a distance equal to the thickness of the book. Starting at page 1, by page 1000 your margins are 2 inches to the right, and if you never adjusted the camera you are cutting off the edge of your text.

Now then, with a movable base that centers itself with every turn of the page, every single page of the book will be perfectly centered and equidistant from the camera. Whether you are photographing page 1, or page 1000, every page will be virtually the same. This will let you do a batch crop with little fuss or manual editing that gets the book finished in the least amount of time.

In a flatbed scanner the pages of the book are pressed against the glass. The page being scanned is always the same distance from the scan head. You end up with pages all the same size, and with the same DPI. With the type of adjustable base in the DIY you can achieve the same results with a digital camera.

Now to improvise the sucker… A DIY version of the DIY.

http://www.roguesci.org/images/BookDriveDIY.jpg

[megalomania]

I partially finished my copystand today. I still have to add the backs to the angled blocks of wood to properly support the book, but that is incidental to the rest of the device. Total cost of this project was under $20.

Here is a basic rundown of how I built it (explicit details to be included in my updated ebook).

I started with a 36 inch Rubbermaid shelf from the hardware store; it is just some black laminated particle board. Cost; about $6 and change.

The slider is a ball bearing drawer slide for a kitchen drawer. Cost was a steep $11. You can get much cheaper drawer slides, but I wanted a ball bearing one for the smoothest slide action.

I cut a 14 inch length of the shelf for the base. The drawer slide is 14 inches long, so that is what determined the size of the base.

The drawer slides were screwed into the base board with the included screws.

I had a spare length of 2x4 that I marked to 40 degree angles with my protractor. I cut the board with a miter box saw I have. With one cut to 40 degrees, I just cut the board straight and this piece was also 40 degrees. The two blocks with angles of 40 degrees actually make a 100 degree angle between them, and this seems to be an optimum angle for holding a book open.

The top board is a 12 inch length of shelf. I wanted to accommodate a book width of 5 inches maximum, and the angled blocks are each 3.5 inches, so this is what determined the top board should be 12 inches.

Using an electric miter saw with a 1/8 inch blade I cut two strips into the center of the top board. This created a ¼ inch slot in the center. I marked off 2 inches from either end to prevent the board from simply being cut in half.

I purchased a ¼ inch double sided screw that has wood thread on one side, and machine thread on the other. I drilled a hole into my angled blocks and then screwed in the wood screw part. This leaves the machine thread part sticking out. No surprise here, the ¼ inch sized screw fits nicely into the ¼ inch groove of the top board. The board is only 5/8 inch thick, and the screw stuck about 1 inch out of the angled block.

I purchased some washers and wing nuts to fasten the angled blocks to the top board. Keeping in mind that the machine threaded part of the screw sticks out about 3/8 of an inch under the board allowed me to place the washer against the bottom, and then secure it tight with the wing nut. The wing nut fits on the machine thread part.

The wing nut sticks out 1/8 inch more than the clearance of the drawer slide gives me, so I had to add the spacer blocks to raise the top board up a bit. I just cut four 1 inch strips from the 2x4, attached them to the drawer slides on the bottom, and used drywall screws to fasten the top board to the spacers. I countersunk the screws into the top board just because it looks slightly better. I used drywall screws because I had them handy.

I actually didn’t drill the holes into the angled blocks myself, so the person that did it did not properly measure where the hole should go. I said screw in the double sided screw near the front, but the screw was so long it may have poked through, so it was placed near the center. This resulted in the unfortunate condition of me being able spread the blocks 4 inches wide instead of 5 inches. I don’t plan on scanning that many books thicker than 4 inches, but if I do, the blocks can be easily modified.

Here are some possible modifications I would consider: I searched the Internet and found some objects called “linear slides.” Some of these are expensive, some are cheap, like $10 each on ebay. These might be better than my drawer slides, they might not. I would have to get one and play with it to see how smoothly they actually slide from side to side.

The end result is what you see in the pictures. Picture 1 is a top view from the front.
http://www.roguesci.org/images/copystand1.jpg

Picture 2 is a top view from the side.
http://www.roguesci.org/images/copystand2.jpg

Picture 3 is a view of the bottom board with the top slid all the way open.
http://www.roguesci.org/images/copystand3.jpg

Picture 4 is a bottom view of the top board showing the wing nuts and washers in the slotted groove.
http://www.roguesci.org/images/copystand4.jpg

Picture 5 is a closeup of the angled block with the washer and wing nut.
http://www.roguesci.org/images/copystand5.jpg

Picture 6 is the finished product with a book in the cradle.
http://www.roguesci.org/images/copystand6.jpg

Yes, I know I didn’t quite cut the block on the line there. It actually does not matter if it is precisely 40 degrees; it just matters that the camera matches the angle of the book. You can see where I penciled in 40 on the block of wood.

Now I have a few more problems to address. I am not even going to address the platen now, that is a complex issue I have not quite decided on what to do yet. Lighting is my next step. My copystand (pictured below) has 4 flanking lights angled at roughly 45 degrees. This is fine when documents are lying flat, but since my books will now be both elevated and angled, I can’t use those lights. The angle of the lights needs to be such that they do not reflect directly into the lens. What I need is a light directly above the book. Essentially the cameras will be where the lights are now, and the lights will be where the cameras is.
http://www.roguesci.org/images/copystand7.jpg

I am somewhat dissatisfied with the quality of lighting I am getting from my copy lights. I think more light is in order, much more light. I wanted some new lights that are cheap and bright, but these two conditions seem to be mutually exclusive. I originally envisioned building an array of 16 incandescent 100W lightbulbs, but it turns out that would A) be far too bright, and B) a circuit breaker can only handle 1400W before tripping. Halogens would literally cook me alive if I used one of those, professional slave strobes would cost a fortune, so that left fluorescent lights.

I was reticent about using fluorescent lighting, so I checked up on some facts. Modern digital cameras can compensate quite well for the color temperature of fluorescent lighting, and there are better quality flicker free fixtures available. I priced some of the overhead fixtures, but only the 4 footers are cheap. Those just seemed too big.

Then I discovered the wonderful world of compact fluorescent lightbulbs. For those of you who may not know what a CF bulb is, those are the twisty “10-year” bulbs that fit a regular light fixture, but are fluorescent bulbs. I always thought these were very expensive, but prices have dropped considerably in the last few years. I bought 14 of them for $3 a pop. Only 6 are for my lighting project, the other 8 are going around the house.

CF bulbs have the added benefit of being flicker free (or flicker less) compared to the long tube bulbs because of the integrated ballast they have. I bought 100 watt “daylight” bulbs, which offers a whiter light comparable to natural sunlight. I bought 6 fixture sockets that I will have to wire up myself. The CF bulbs are actually 23 watts each; they put out the same about of light as a 100W incandescent bulb.

The 140W of the CF bulbs will not blow a fuse, but I still get 600W of light output, which is 150% more light than I am using now. I plan on wiring the lights sideways, using a reflector above the lights, and a translucent diffuser below. The diffuser is my fancy way of saying I will be using a ceiling light :) http://www.homedepot.com/cmc_upload/HDUS/EN_US/asset/images/eplus/162473_4.jpg

[Maitreya]

Copying a book is easy. Getting rid of faded edges is complicated. Stopping glare from happening is hell.

Megalomania, do you think this "DIY of the DIY" will reduce glare from books with glossy pages?

[megalomania]

No, glare from glossy pages is a very persistent problem. Getting the right lighting angle can reduce it, and even eliminate it. Photographers have their tricks for photographing reflective items, like gems for example. I think the lightbox, or light tent, is the prevailing solution.

[megalomania]

Atiz, the book scanner company, released a new more affordable book scanner in November of 2007. The BookSnap retails for $1600, not including the two digital cameras required to work the machine. This book scanner is intended to be affordable by average consumers. Many might think $1600 is a bit steep for something like this considering you still have to supply the cameras, but I think it is a considerable improvement over any existing system. The next most affordable system is the BookDrive DIY at $4500, again sans cameras. Atiz is marketing the BookSnap as a book ripper, reminiscent of DVD ripping, not the damaging kind of ripping.

The price is still a bit high considering it is a fancy camera stand with sliding clear platen and book cradle, with some lights thrown in. Engineering wise it looks quite sturdy and well built, but $400 would be a better value. The BookSnap is manually operated and it looks to be essentially a cheaper more refined version of the BookDrive.

I regret that I have been slacking off with building my own version of the BookDrive. I bought my linear slide months ago, I have the lights half wired in the lab, and I have all the brackets and lumber for the frame. I just haven’t bothered to put the damn thing together. I never did hear back from the plastic vendor about how much the platen would cost me, so I guess that’s where I have been stuck, but I could still build the rest. I’m not really mechanically inclined, and I don’t have all that many tools of my own, so progress has been slow.

I found another website in October where a guy posted about his DIY book cradle and experiences with using a digital camera to scan books. His methods and solutions read very much like my own. His cradle looks much nicer than mine, but it is not a sliding one like I built: http://www.mobileread.com/forums/showthread.php?t=13848

Atiz has a very nice animated gif illustrating the importance of using a moveable cradle. A picture is worth a thousand words, and the image explains the process better than I could with ten pages of writing.
http://booksnap.atiz.com/image/how_to_scan/18.gif

Momentum is gathering in the DIY book scanning movement. Software appz are still lagging behind unfortunately. In another few years people will be scanning everything they can get their hands on and putting the results on the Internet. I foresee libraries and universities buying these and charging patrons a small fee to scan books, and then the library will keep the digital copy in its archive. Free labor that way… I also foresee the death of printed textbooks as students scan and exchange books in droves prompting publishers to force universities to charge mandatory fees for every student enrolled in a class. The publishers will make even more staggering profits by forcing everyone to buy electronic editions, which cost just as much as print editions, but which cannot be sold back or transferred. Publishers will mandate outrageous prices and eliminate the used book trade altogether. Students will pirate the ebooks initially, but publishers will extort the money directly from the universities, who will in turn charge a textbook fee as part of the cost of tuition. Student loans and grants will end up subsidizing much of this cost, and the taxpayer ends up making the publishers even more obscenely wealthy than they are now.

This will happen with or without piracy. They are greedy, ebooks are more profitable, and they have no compunction about using unethical and illegal means to steal your money. The phantom specter of book piracy is just a smokescreen to lull the brainwashed sheep into towing the “copyrights are good” line. Indeed only with piracy do we have a chance of avoiding this bleak future. If people stop paying money for books, the publishing industry will collapse and be forced to correct its vile ways. Without the money to bribe lawmakers into passing copyright laws favorable only to publishers, the will of the people might just be heard and the current criminal copyrights laws be repealed.

But I digress… Hopefully very soon I will get my act together and finish building my book scanner. It may not look pretty, but it should do exactly what the BookDrive and BookSnap do for less than $100. I do wish now I would have kept better track of my expenses on this thing. The linear slide cost about $45, the wood, brackets, and drawer slider about $20, the light sockets and wire nuts less than $10 (special thanks to a maintenance worker who generously donated the wires on behalf of my former employer :D ), some black fabric cost me a few dollars, and the compact fluorescent light bulbs were about $4 each for 7 bulbs. I bought 7 bulbs originally, but for the house, not the camera. Two bulbs are already broken by accident. I originally planned to use 5 bulbs, but now I am up to 7. This may prove to be overkill, we shall see. That’s another $30 for bulbs. I built a platen out of clear acrylic and wooden dowels, but it was flimsy, so I scrapped it. I ended up scratching the plastic which makes it worthless now. There is another $12 down the drain. Who knows how much it will cost me to have a plastics company custom cut and bend a new platen, $50 I imagine, maybe $25. I might go with glass instead, the additional weight flattens book pages better. I need to build a frame and slider for the platen too. I hope to recycle the drawer slide for this now that I have a linear slide. I already had a tripod head, some 2x4 lumber, and a dowel rod for the camera mount, so that didn’t me anything. I spent a few dollars on various screws, nuts, washers, and threaded rod for the camera mount.

As you see from the list of expenses it will end up costing about $150 to build this thing from scratch, most of which is the linear slide and lightbulbs. I got the linear slide used from ebay, it was probably $200 originally. The average selling price of linear slides on ebay is $50 for the size I wanted. There are cheaper ones, around $30, but those have a very short length of travel. I wanted enough length of travel to accommodate thick books with a margin of safety to boot, so I bought a 12” slide. My minimum requirement was 9” but you take what you can get on ebay.

Then there is the cost of the camera equipment. I wasted more money than I would have liked on camera equipment, but at least I can use it for family photos and the like. I wouldn’t mind photographing some nudes… Since most people have digital cameras already, it’s not quite fair to factor in the camera cost for a camera scanner. Still, they are a very expensive component, especially if you use two at a time.

Now that I have added all this up I am a little bit discouraged. What kind of nut would spend this kind of money just to scan some books? I have to keep my eye on the prize. This method saves considerable time vs flatbed scanners, the camera costs are dropping all the time, and compared to the staggering cost of textbooks this method can save you tons of money. A cheaper alternative to the expensive linear slides, and perhaps a ready made light set, will cut a lot of the cost.

[Charles Owlen Picket]

No brown nosing shit here...this seriously is one of the better and more productive presentations I have seen. I have a keen interest in this topic (as you know) and this was very well thought out. I think we share a similar love.

I think I'm going to make the plunge and start one myself. I have the time (or will quite soon) and this presentation just got me hooked!

What's more, I totally agree that ebook profiteering is a joke. But we all saw it coming as people don't treasure the book-makers art nor for that matter does the public read. It's a student's domain and student's are ripe for the picking. ...Now, when someone makes a "notebook-laptop concept" designed to "read" or present a pdf file in a book-like manner and sell that for under $20.....that will make a mint!

[FUTI]

Mega I wasn't visiting forum in a long time and since that polite warning about lack of posting in a long period on top of page caused a sense of guilt I remembered this post about planetary camera. I think you had problems with lights and wanted to make some standardized set of bulbs set around the "gadget" shown on that photo. I'm just curious did you take into consideration "angel eyes";) - CCFL rings that I think you could place around objective of that camera. I guess that incoming angle of light to paper can be important on those glossy paper some books are printed (and the way I see it the sharper the angle light-page-camera lens the better) and also CCFL looks more diffuse to me not to mention that it use less power for the same light output. I see those for sale on the net but alas I live in a shithole country inside the shoebox with internet.

[megalomania]

I purchased a ring light on ebay for my camera when I bought it, but I got ripped off. Anyone who sells macro ring lights for Canon Rebel cammers is a scammer, but not one that technically violates the rules. They sell these in the digital camera sections, advertise them for Rebels, but they fail to mention they do not work with DIGITAL rebels, only the film cameras. Who buys a film camera? The light works, it flashes, but the electronics are incompatible with digital Rebels.

I am in the bad habit of letting a package sit around for a few weeks, sometimes months, before I open it depending on the item. While I did open the package, I did not have immediate need of the light, so I never tested. I only found out some time later, after any reasonable return date, that my light was useless and I had been scammed.

This waste of $150 so disgusted me that I will never buy anything off ebay that I can't get at a store on online vendor. You often have no recourse to return crap to ebay sellers, and almost all of them are scammers. Those bastards (sellers, not ebay itself) need to be held to the same accountability as a retail store if they are going to be "power venders" or whatever they call themselves. They are nothing but fly by night scammers.

A ringlight that would work with my Canon Rebel XT would actually cost $250-$300, so screw that. This is why I decided to make my own light. My current design, a parody of the Atiz BookDrive DIY, does not require the lights to be near the lens.

If I could draw a picture in Inventor Pro (that's another question coming up) I could easily show you in simple terms, but I can't draw, or use drawing software, so I will use the thousand words :)

Book pages are too reflective, using a glass platen makes the reflection unbearable. Photographing glass is tricky because of reflections, but there are tricks. You cannot have the light firing down at the surface being photographed with the same, or close, angle that the lens is. This means a ring light is worthless. If you do not use a glass cover, and you do not photography shiny book pages (most new books are shiny, color books anr shiny, magazines are shiny, old books are not shiny) you can use a ringlight.

To avoid a reflection off of the glass the angle of the light has to be 30, 40, 50 degrees or more away from the lens of the camera. My setup has the light directly above the book, the camera is off to the side at a 50 degree angle or something like that...

I can't describe this in words, you just have to see it to understand. I am a visual person, I need a picture...

[Charles Owlen Picket]

I still shoot 35mm film. I wish I'd have known I would have said something. I even have a Cannon. - You have to REALLY be sharp with eBay and shit like that; they really WILL sell shit - insinuating that it's workable for both types, etc. One MAJOR reason why I still shoot film is the volume of lens I have. You have to check everything for cameras....(But also having a darkroom is a good method to maintain a quantity of chemicals as well.)

What's more...IT WAS AN EXPERIENCE EXACTLY LIKE THAT THAT CAUSED ME TO CHUCK eBay. I really DON'T buy off that thing anymore. If it was me - I'd try to get my $ back. I hate it when they don't add one damn line like "works for film (or digital)" in the ad.

[megalomania]

Here is a schematic picture of the BookDrive (see attached image). Notice how the camera is angled, being eactly perpendicular to the surface of the page+glass, but the lights are in the center (not actually visible in this picture).

My book cradle holds the book open at a 100 degree angle, so the camera is angled 50 degrees with respect to the horizontal surface and the lights. At this angle reflections are minimized, at least I hope they are since I am not done building the platen.
When I used lights on my basic setup I had to angle them to keep reflections from bouncing into the lens. The camera picks up a bright blob, the reflection of the bulb, the gets less and less the greater the angle, until it no longer covers the text. My new lights will have a diffuser to minimize and point reflections that obscure text. The diffuser also evens out the illumination on the page. I had a hell of a time editing pages when I first started doing this because of uneven illumination.

I am actually almost done with my lights, by the way. They are fully wired, everything works! I now have a few finishing touches to add. I have to get more CF bulbs because I don't have enough anymore. The bulbs I bought some time ago are now being used to light my bedroom, hallways, bathroom... :)

Right now I need to add hooks to the lights so I can hang it from the as yet unbuilt stand, I need to build the diffuser cover, and I need to get new CF bulbs.

I did a little more research on using compact fluorescent lights for photography work recently. Photographers who use the lights recommend a color temperature of 5000K, which corresponds to "daylight" light. I didn't see anyone suggest using the 6400K bulbs, but I didn't see anyone discourage their use either. Most CF bulbs you get at the hardware are the soft light, which needs color correction. The 6400K lights are a little on the blue side and might need corrected. 5000K is supposed to be closer to white, and not require any color correction.

I am going to try the Litetronics Neolite brand of bulbs. They are a bit more expensive than hardware store brands (usually GE), but I can't find higher wattage daylight CF bulbs in my area. The Neolite brand sem to be of better quality, they have the lowest mercury content of any CF bulb (I hate mercury on general principle), and they have the highest lumen output per watt. The cheapo hardware store bulbs seem to burn out if turned on and off too much, have inconsistent light output per bulb, have a less accurate color spectrum, and in general suck. Since this is photography, not my living room, I want to get something a little better. The Neolites are actually only about $1.50-$2 more per bulb vs what I can buy at the store (factoring in shipping vs tax).

I have been racking my brain recently trying to figure out what to use as a diffuser. Professional photography umbrellas cost a fortune considering they are just umbrellas with nylon sheet over the front. I could buy an umbrella, but the thought of cutting and modifying all those wires seemed daunting. I next thought I could get a metal bowl, which is both reflective and heat dissipative, but when I priced them at Wal Mart the big ones were rather expensive. A plastic bowl might hold in too much heat. I don’t think there really is a bowl large enough, besides I didn’t relish the thought of cutting off the bottom of a stainless steel bowl anyway. Then I thought why not use a lampshade? I checked out some lampshades, which are all too small and very expensive, but I can easily build my own.

With an old wire hanger or two I will form the basic shape of my parabolic reflector. Hangers, like lampshades, use a stiff wire that will hold it shape. I could buy a lampshade and reshape it, but I didn’t find any black ones where I could reuse the material. I will go to Goodwill or something and get a nice black fabric to cover the reflector. I think a heavier fabric, like the kind used on lampshades, would offer better heat resistance, and be breathable to let heat out. Or rather I should say my mom will be getting the fabric and stitching the reflector since I am drafting her for this part of the machine. She likes sowing and craft stuff. I don’t have even so much as a needle, let alone a sowing machine, or the skill to use it.

Once the light reflector is done I will cover the inside with aluminum foil. I read one photographers caution that most of the light from CF bulbs goes sideways, which seems accurate in my experience. Since my bulbs point straight down, a reflective surface on the inside can’t hurt. Now I only have to figure out what material I will use as the actual diffuser. I was thinking about several layers of tissue paper. Something like a pillowcase or bed sheet might be a little loose in the weave letting pinpricks of light through. The area to be covered is too large to use milk jug plastic without taping and gluing bits together (creating shadows). I could get some nylon fabric like the photography diffusers use, but I don’t know if a fabric store carries such a thing (me being new to the whole fabric domain, I never took mental stock of what is available at such stores). Plastic bags are too thin (not enough diffusion) and crinkly (introduces shadows). A white plastic garbage might work… I will start with the tissue paper first.

I decided to skip having a plastic company cut and bend some acrylic plastic for my platen. I think glass has a superior weight and heft to flatten book pages better compared to plastic. The heavier platen should also make the cradle slide better when it is lowered. I don’t really like the ant-glare glass I have for my basic setup, so I will stick with plain glass. I would love to get some TruVue Museaum glass, or another kind of anti-reflective glass, but that stuff costs a fortune. Maybe someday. For now I bought a cheapo picture frame on clearance, but they only had the one, and I need two. The frame is 11x14, the size I want my platen to be, and it has a black aluminum frame. I can use the black back matting of the picture frame as part of my cradle, which saves me having to buy black posterboard. The aluminum frame solves the problem of me making a frame for the glass too. I will have to modify the frame to suit my purposes. The platen is really the hard part in the whole setup in my opinion. I just hope I can find another cheap frame like I have. My frame cost me $3, the closest thing at Wal Mart was $12 :(

[Charles Owlen Picket]

This is good stuff. I'm very impressed with the thumbnail (if that's the product of Inventor '08 - I'm downloading that thing ASAP!). I would also try tissue first.
.....I also remember an older enlarger that used very light, transparent frosted glass. It was repaired and someone turned the glass backward. I always though it a pain in the ass as I was concerned not to smudge it when younger. But in this application, high quality frosted glass turned to minimize reflection may be ideal. Obtaining it would be a chore however.

As you have an issue with mercury, I have with acrylic. Anytime I tried to work with it the results were much less than satisfactory. If you don't have to use it; don't. If you want to take a crack at some shortcuts, find a big city camera store and see if they have older darkroom equipment. They may not only have transparent frosted glass (that is actually FOR the application of minimizing refection) but may have some equipment that could be cannibalized. Some enlargers are quite large!

[Just a thought] Some of the professional level shit back in the 70's - 80's would surprise you at their size. For the type of pictures you want they may have parts you couldn't find anywhere else. ...But didn't you try cannibalizing stuff when you first were thinking about this idea? If so; I apologize for re-hashing this issue.

[megalomania]

I went to a local glass shop to get a pane of anti-glare glass a couple of years ago for my basic photo copy stand. This glass has the glazed/frosted front. I don't like how the glass makes the pages look slightly fuzzy. You can only see through the glass when it is pressed directly against paper, but even then it is just a bit hazy. I will stick to clear glass.

The use of a diffuser should eliminate any washouts from a bright reflection, and keeping the lights ~50 degrees from the lens should limit reflections.

Should... I have not tested the full system yet. I still have a few finishing touches to add to the lights, like a way to hang it up.

Also, I wonder if someone can help me with a screw problem. My linear slide needs 3 mm diameter machine screws, a minimum of 7/8 inches long (~23 mm). I need 4 of them. I went to the hardware store (Home Depot), but the smallest they have is 4 mm. I have one 3mm screw, but it is only a cm long. Anyone know a cheap place to get such screws?

[justme]

Mega, you can get a whole bunch o' screws at www.mcmastercarr.com that will fill your need. I did a quick search, narrowing it a bit, and came up about a dozen that would fit your description, assuming you don't want plastic. The cheapest is item number 92005A130 which is $2.85/100. McMaster Carr is probably one of the tinkerer's best friends. They are not the cheapest, but it is about the best one-stop-shopping for hard to find parts and materials.

[megalomania]

Oh, I never thought about them. I have the pipe and hose catalogs, a few others with the large stuff, but I don't have the whole set. McMaster-Carr has, what, 20-30 different catalogs all together? I was hoping to find someplace local to walk in and grab some, but I don't suppose there is much of a market for specialty screws if not even the big box stores carry the stuff.

EDIT: I found what I need, #91420A128 machine screw, metric size M3, Phillips head, 20 mm long, steel, zinc-plated, 100 pack for $2.80. That's 196 more than I need, but maybe they will come in handy someday. The hardware store charges $0.30-$0.60 each for the screws they have.

[justme]

I always thought that it was difficult to get too many odd screws. You never know when you might not have one.

[megalomania]

Due to complications with my original cradle design I came up with a completely new and improved cradle.

Here is my rough assembly in SolidWorks so far:
http://www.roguesci.org/images-forum_posts/book-digitizer.gif

The problem I was faced with was how to access the butterfly screws that tightens the adjustable blocks of wood that hold the book. The drawer slide lets me pull the whole thing out from the table so they can be accessed, but my new linear slide will not let me do this.

Also, with the slide in the center the distance the blocks can travel is limited. I only have the one linear slide, it is extremely expensive, it puts the whole project WAY over budget for it alone. Getting a second one would be wonderful, but it is NOT an option. Not for me, but for other people when I make this design available to the public.

I was playing around with a few of the right angle brackets I bought to connect the 2x4 posts to the baseboard when I realized I could use these as guides with a very limited clearance from the cradle board.

As you may see in the picture the cradle now has angled 2x4 boards going all thge way across the cradle board. This is a better improvement because I never liked how small my blocks of wood were, barely 1 inch across is insufficient.

The brackets will need a hole drilled in the side and a nut soldered on so I can add a knob to use as a tightener. The brackets will hold the wood book supports on both sides keeping the supports straight. The linear slide elevates the cradle board off of the base board by about 13 mm, whereas the thickness of the bracket metal is only 1 mm. The brackets can slide freely across the cradle board, and can even be removed if need be, to accommodate a variable book thickness of any conceivable size up to about 9 inches.

I also found an alternative to using a linear slide that I will design in theory since I can't buy the parts now. Perhaps in a few years I will build version 2.0 when I get more money. The new theoretical design uses linear bearings.

McMaster-Carr sells 1/8 inch linear bearings for just under $14 each. I found several 4 packs of 20 mm linear bearings online for about $25. The design does not save money overall (about $50-$60 overall for a ebay linear slide or for 4 linear bearings) but it allows for a 2 rail system.

I would prefer to have 2 linear slides, just like I have 2 drawer slides in the cradle pic in my earlier post. Linear bearings are much cheaper. I don't think one needs the special precision machined and specially hardened metal rods that are supposed to go with the bearings (about $10/foot depending on the thickness), nor does one need the special end caps.

Assuming you can get a cheap four pack of 1/8 or 1/4 inch linear bearings from ebay or google checkout, one should be able to get regular steel or aluminum rod at the hardware store to use as the rail for the bearings. The rods can be supported by blocks of wood at either end, the bearings are then affixed somehow to the cradle board.

Using two support rails and four linear slides, one at each corner, makes for a more stable cradle. I only have the one linear slide with a very small top surface to affix the cradle to. Ideally all the weight of the book should press right in the center, but if I accidentally press my hand down on the side of the cradle the lever force could snap the wood, or even break the linear slide.

Here are a few pictures I found online that sort of show what I am talking about. Notice each pic includes a pair of metal rods suported at each end by a side piece that elevates a top plate from a bottom plate.

[megalomania]

As for the platen... I gave up trying to find a suitable picture frame. Picture frames are either too expensive, the wrong size, or just flimsy metal. I also could not find any suitable aluminum U channel bent with a 1/8 inch gap (to fit a 1/8 inch pane of glass inside).

Getting a custom fabricated frame sounds too expensive and time consuming. I'm supposed to call up every metal shop in the tristate area and ask if they will waste their time bending a few pieces of metal for me for next to nothing? It would cost a fortune to have a picture framing shop do the job.

I found a very nice solution to this problem. Aluminum channel only comes in a few different configurations depending on the thickness. I want a nice sturdy and heavy 1/8 inch thick metal, but the smallest I can get the inner bend is 1/4 inch. I want to use 1/8 inch thick glass, so what to do about the extra 1/8 inch?

Rubber liner. McMaster-Carr sells a 1/16 inch thick rubber liner meant to fit inside U channel. Each leg length is 1/4 inch, the size U channel I want. The rubber liner goes inside the U channel and the glass fits inside the rubber. The top and bottom thicknesses of rubber, each 1/16 inch, times two is 1/8 inch. An 8th inch of rubber and an 8th inch of glass fit perfectly in a quarter inch of U channel.

Using the rubber is actually a good thing I suspect because this should reduce stress and vibration on the glass from the constant raising and lowering of the platen. The glass will be pressed hard against the book when it is lowered, so rubber padding should protect the glass somewhat.

The rubber is only something like $0.10 per foot.

I figured I would need to add some sort of glue or caulk to seal the glass to the frame anyway, so using the rubber means I don't necessarily have to permanently connect the glass to the frame.

[megalomania]

I completed the design of the Book Digitizer in SolidWorks last week. While this may not be the ultimate final design, it is what I like to call the "release candidate 1" design.

I have some new design ideas for how the light is held up. I don't like the 2x4 arrangement, but I just added that for lack of something better at the time. I think I will use a hanging lamp type arrangement. A threaded pipe attached to a flange, a 90 degree elbow and a short pipe sticking out, the light hangs from that by wire or chain. The flange gets installed in the back of the base board in the center.

I will still have to measure that because the diffuser might require the light to stick out so far that it over shoots the book cradle. I may end up using an actual floor lamp that is separate from the Book Digitizer so that it can be easily moved and positioned as necessary.

Anyway, here are a few pictures of RC1 of the Book Digitizer. You can't see all of the components since this is just a screen shot from one angle. There are CF bulbs and sockets inside the light that are hidden by the diffuser. That is not the shape of the diffuser I want, I could not figure how to model a parabolic shape in SolidWorks without spending many hours learning how.

I designed this in four separate assemblies; the base, cradle, platen, and light are all their own assembly files, and then I made the final assembly by bringing them all together.

As soon as I can figure out how to export everything with the dimensions I will show a pic with those. All of the component sizes are as close to what they will be in the actual build as I can get them. The point of me using SolidWorks is to figure out what all the lengths and angles need to be.

I calculated a few important parameters that are specific to my camera and lens. I determined the distance from the lens to paper required to exactly frame a 8.5 x 11 page is 30.5 inches. This drops to 20 inches to frame a 5 x 9 page. The dowel rod that the camera slides back and forth on is 18 inches long, or rather it can slide 18 inches before hitting the support rods.

The "zero point" as I like to call it is when the support rods holding the dowel rod are as low as they can go. The camera needs to be raised and lowered so that the center point of the lens is exactly perpendicular with the center of the page being photographed...

Just to interject, a camera's aspect ratio means it close crops the 8.5 inch part of a 8.5 x 11 inch paper, but the 11 inch part actually has some additional area. Since cameras photograph things as a rectangle, and book pages are almost always rectangular, books are photographed sideways. The only dimension that a camera needs to worry about is the width of a book, so I will only refer to that dimension...

As I was saying, a small 5 inch book has the middle at 2.5 inches. A regular 8 inch book has the center at 4 inches. The largest books I want to be able to photograph are 14 inches, and their center is at 7 inches. The camera only needs to be raised 4.5 inches from the smallest book to the largest, so the rods that support the dowel do not need to be that long. I made mine about 8 inches just to give some leeway, and one must factor how far into the support arm the rods need to go as well. I already bought a 2 foot threaded rod thinking I would need to raise this much higher. SolidWorks helped me see a solution to this problem quite clearly.

As a consequence of the camera only needing to rise a few inches from one extreme to another I had to considerably shorten the height of the side arms that hold the camera support arm and have the slides attached. Considering the dowel rod, tripod attachment, and the height of the camera adds up to exactly 4 inches to the center of the lens, I had to shorten the side arms accordingly. I may need to get a shorter drawer slide.

Speaking of drawer slides, I decided to try these out before resorting to using a very expensive linear slide. Unfortunately I already bought my linear slide, but I can always sell it right back on ebay for what I paid for it. I tested my drawer slide under a load, and it moved much easier. I discovered this on my linear slide and I wondered if the drawer slide would do this too. Both slides seems to jerk and need extra force when there is no weight pressing on them, but with a little weight they slide smoother.

Attached is the image of RC1. I found the assembly of the camera and the drawer slides in a free parts library. The camera is actual a Minolta Dimage point-and-shoot, not a Canon Rebel SLR like I have. I could not find a model of any other cameras besides this one, not that it really matters because it is just for show. Almost any type of digital camera can be used on the Book Digitizer.

[monkeyboy]

Just ran across this over on avax, thought somebody here might find it useful.
I haven't tried it yet, myself.

Book Restorer 4.2.1.0
OS: Win32 | 10.2 MB

Book Restorer™ software is the ideal tool for restoring images because of the diversity and quality of its processes.

Book Restorer™ includes numerous features entirely dedicated to the digitization of materials such as books, incunabula, newspapers, registers, property books and maps.

Book Restorer™ is the ideal application for managing, restoring, storing and publishing written works.

Book Restorer™ lets you display and view all the images in a book digitized with a scanner.

During the import and export phases, you can select any of the following formats to optimize your files for their intended use: TIFF, JPEG, BMP, PDF or PNG.

The Book Restorer™ architecture allows you to simulate a book structure.

The automation module enables you to run restore and export processing as a background task.

Book Restorer™ is “must-have” software for anyone who wants to restore or even improve the quality and accuracy of original digital documents.

Assistants
Book Restorer™ provides several assistants to aid users in quickly familiarizing themselves with the tool and mastering its use. These assistants provide vital help in the essential steps of using Book Restorer™.

Modules
Book Restorer's™ Automate module (BKR Automation) lets you import, restore and publish your images in batch mode. This is of particular benefit in the case of documents containing numerous pages of similar appearance. These three tasks can be started simultaneously, allowing you to avoid the entire phase of image importing and script programming.

In contrast to the Automate module, Book Restorer's™ Watcher module (BKR Watcher) lets you process images as they are being inserted into a folder.

With Book Restorer's™ Quality control module (BKR Control), you can create a report on the restoration processing performed on the images of a book created with Book Restorer™. A quality coefficient is calculated for each book image and each restoration applied, which makes it easy to define which images should be verified and inspected.

Image processing
Book Restorer™ includes many image processing modules. It is also easy to integrate new ones.




Rapidshare
http://rapidshare.com/files/118657289/BkRstrr4210.rar

Depositfiles
http://depositfiles.com/files/5677093

[megalomania]

Oh yes, I was going to mention this but Rogue Science was down then. I actually was idily surfing Avax the very minute this app was offered! The date of the release was the day after I got it, talk about zero day.

I tested it in Sandboxie (which I highly recommend), no viruses detected and it works good. Book Restorer still has the annoying crash bug when trying to open photoshop, even though it lasted longer than usual. I have yet to incorporate photoshop actions with Book Restorer since BR does a well enough job. Photoshop processing is best done before using BR (batch rotating, cropping, deskewing). Book Restorer has a particular dislike for jpg files from a digital camera, as many as 25% or so fail to import. Photoshop can help batch convert all images to tiff, the preferred format of Book Restorer.

I am trying to write the "manual" on Book Restorer. It is poorly documented, and it is one of the most bug riddled pieces of software I have ever had the misfortune of using, and at the same time it is one of the most useful and powerful programs for what it does. Basically I can only document the program through trial and error of the features.

I got a price quote a few years ago. Sit down for this. The full package with tech support is $14,000, and just the software itself is about $6,000 - $8,000. If you can pony up for this you will want (need) the tech support.

If I knew anything about software coding, I would love to make something to use Book Restorer as a backend and code my own better front end. Particularly frustrating is the softwares lack of an undo feature. It has the entry in windows taskbar, but it is perpetually grayed out. The developers say the feature is not implemented...

[megalomania]

The build of the machine is currently underway. For the last couple of weeks I have gathered the last bits of miscellaneous components and tweaked my Solidworks design. I finally brought together the borrowed tools I needed, the compound miter saw, drill press, big clamps, etc.

So there I am, measuring thrice and cutting once, but real world applications do not always meet design specifications. The design is fine, it's the wood that is a problem. I hate to say it, but even after spending 20 minutes looking for the straightest lumber I could find, the boards are still a little off.

Now it is back to the drawing board for a little while. The dimensions and calculations of the required widths, heights, and angles are still fine, but I need a straighter material. Unfortunately, that means metal, which costs more and requires different tools to work with.

Providence has been kind to me recently in that I found a tree stand growing in my back yard, chained to one of my trees. This makes the 5th one in as many years. I would like to know who plants those fine metal trees… After spending a few hours dragging that heavy ass unwieldy chunk of metal through dense foliage, which for the last couple of weeks has been repurposed as a makeshift bridge over my stream, I finally got it back to my workshop. Thank the gods it breaks down into sections.

It did occur to me on the way back through the woods, as the day's promised thunderstorm finally brewed up around me, that I was carrying a giant lightning rod on my shoulders. Oh the great lengths we go to for our art :)

I would prefer aluminum as it is lighter and easier to work with (cut and drill), but I can't beat the price. Now I begin the process of redesigning the machine to utilize the metal. My redesign must incorporate the overall dimensions of the tree stand because it already has welded cross members (it’s a ladder).

Also, while I was building the machine with wood it occurred to me to be more mindful of portability. I already factored in using bolts instead of screws, wing nuts here and there, but I still designed some permanent attachments that would have made the stand unwieldy. This was an easy change, but that does not matter now.

With a metal stand, unless I weld the thing together, it will have to be bolted to assemble it. I don't plan on taking this thing on the road, sneaking it into the Library of Congress or some such place, but it would be nice to break it down and stow it in a compact manner when not in use.

I don't suppose anyone can tell me the difference between metal and wood drill bits? I picked up a few grab bags this February, and now I don't seem to recollect which bits do what. Is there a way to tell by sight, color, twist, or does it matter? This is another reason why I would prefer to use aluminum, if I use the wrong type of bit at least it will not get too fraked. I don't suppose my wood bits will last too long in steel, though.

[monkeyboy]

I'm very interested in how it comes out. Keep us posted Mega.
I just picked up a pair of full extension commercial drawer slides, with ball bearings.
I'm thinking about either metal or (I think it's called) UHMWPE, a company around here makes stuff out of it, so scraps are pretty easy to come by. It's a really heavy, durable, machinable white plastic...

Drill bits:
Basically, if it looks like a standard twist drill, it's probably for metal, but can be used in wood. Wood won't hurt it. If it has an odd looking pilot tip, it's probably for wood only.

Technically, general purpose twist drill bits (wood OR metal) have a 118° tip. While a sharper angle, up to around 90°, is for soft materials, like wood or plastic. And a flatter angle up to around 150° is for harder metals.

Obviously cheapo drill bits are going to be softer, so while they'll work fine in wood, any kind of decent metal is going to destroy them.

[sbovisjb1]

My Uncle is a photographer and He told me that he uses the Umbrellas to reflect light so there is no glare. He said anything white could work, white ceiling, wall, board, etc.
http://en.wikipedia.org/wiki/Reflector_(photography)

I'm quite excited about this project. Now to make this automated so that it wont take forever. :D At the university, they have full time bookbinders. If they want to scan a book, they unbound it, and place it into a office photocopier/scanner where the pages just slide in and all you do is take them out and rebind them.

[megalomania]

I was going with diffusers for a while, but I had trouble designing something good enough to cover the light I designed. I realized my problem was the hexagonal base I made for my lights. This made it far more challenging to build something to fit on the sucker.

I kept going through material after material. I built, and scrapped a wooden frame held by wire. I used tissue paper, sheets, tried aluminum and Mylar internal reflectors. I checked out all the DIY photography sites about how they made the things. I got to the point one day this summer where I spent 18 hours straight without eating, taking a break, or moving from my chair at all except to pee a few times, trying to model a damn pyramid in stupid SolidWorks for my light shade diffuser.

After that sobering incident I said fuck it, and ended up buying a cheap ceiling light! It has a nice translucent white glass dome diffuser and a wide base to accommodate additional bulbs. I still need to mod it to put the extra bulbs in. I will probably replace the existing sheet metal that the sockets attach to with one of my own that has additional sockets.

Indeed I have wasted a lot of time, a little money, and I have a bunch of designs for the same thing I am unsatisfied with. I now know I can’t drill a perfectly straight hole worth a damn, that 2x4 lumber is not very straight at all, that SolidWorks is really powerful once you get to know even a few tricks, how to properly wire lights and switches, what the difference between machine and wood screws are, and where knowing the trigonometry trick SOH CAH TOA actually has a practical application.

So here I am back at square one trying to redesign a better prototype using the lessons I learned along the way…

[megalomania]

Maybe I should have asked this some time ago, but now I need your help. I have very limited experience with woodworking and metalworking, so I don’t have a very tight grasp on what I could do, or how hard what I am doing is.

We have learned I can’t drill a perfectly straight hole in wood using a drill guide. The tolerances on my old design were extremely high, nothing short of absolute perfection would work. I can’t get that with a drill guide, and the parts were too big to fit in my drill press. The wood is also slightly warped, my bolts were a bit too big, and I am limited to the parts I can get from the hardware store (for the sake of keeping costs as low as practical).

Wood is all well and good, but I was never satisfied with the sturdiness of the structure. I decided, based on me having some scrap steel square tubing, to use steel square tubing. I can’t beat free. I would prefer to use aluminum tubing, but at $2 a foot vs $0 per foot, you do the math. [I don’t remember the exact price of the aluminum; it differs by length and wall thickness]
The steel is in the form of a tree stand I found while walking in my back yard. Some kind hunter planted the stand seeds and it grew up into a 15 foot steel ladder with vines of chain padlocked to my tree. As much as I enjoy getting shot at by trespassing hunters on my own land, I decided to harvest the stand and repurpose it.

I modeled two ladder rungs to use as the base since it already has the steel welded together. The dimensions are indicated in the first attached illustration. My goal is to use as much of the stand as I can without having to mod it much. I am not sure (yet) as to what the minimum width and height of the stand should be, but the 28.75 inches of two ladder rungs should be more than enough width for the base.

I tried to model what the minimum and maximum parameters of the stand need to be. I have to set some limits on what can be practically photographed to keep the size of the MegaBook Scanner reasonable. I assume the following:
MAX book thickness: 4 inches
MAX book width: 8.5 inches
Angle of open book: 100 degrees

This means I am designing my stand to accommodate a book as large as 4 inches thick and as wide as 8.5 inches not including the margins (actual book width would be about 10 inches or more). There are standard sizes of books for width and height if you care to look them up, I have. Most books are 6 to 8 inches wide, at least all the ones I would want to scan. Anything bigger than an 8 inch wide book is quite rare, and to digitize would exponentially increase the size of the stand. I don’t want a map scanner here.

A 4 inch thick book needs to move 4 inches from right to left when photographing. The stand needs to be wide enough to accommodate an 8.5 inch wide book, opened to a 100 degree angle, with 4 inches of space to move, and enough extra space to fit the glass of the platen.
If my math is correct, the length of an obtuse isosceles triangle is given by c2 = a2 + b2 -2ab cos(gamma), where a,b, and c are the opposite, hypotenuse, and adjacent sides of the triangle and the angle gamma is the angle between sides a and b.

The angle gamma (at C) is 100 degrees for my books. The angles alpha and beta are both 40 degrees. The largest book is 8.5 inches, which makes the sides of the triangle a and b 8.5 inches. The equation is c2 = 8.52 + 8.52 - 2(8.5)(8.5)cos(100) = 144.5 -(-25.09) = 169.59
Then c = sqrt 169.59 = 13.02, or about 13 inches wide.

The book is actually 13 inches wide opened to the first page, but the thickness of the book also has to be factored in. The book will form a right triangle with side b (adjacent) equal to the thickness of the book. The angle A of this triangle is equal to 1/2 the angle of the book, or 50 degrees. Angle A is 50 degrees, length of the adjacent side is equal to the thickness of the book (4 inches max in my case). Using the good ol SOH CAH TOA formula, rearranged to get: hypotenuse = (adjacent)/(cos A) = (4)/(cos50) = 6.22 inches.

If we consider the top right hand inside of the book to be the peak of the triangle, it bisects our 6.22 inch triangle in half. This means our 4 inch book sticks out only 3.11 inches. The maximum width of a 4 inch thick book measuring 8.5 inches by 11 inches is 13.02 + 3.11 = 16.13 inches, or about 16 and 1/8 inches when opened to 100 degrees.

All this math tells me my base has to be a MINIMUM of 16 inches wide just to fit the book inside. I still need to be able to move the book 4 inches from side to side. I have never seen that many 4 inch thick books, and the thickest I have is 3 inches, so there is a little extra breathing room factored in here. The platen should be somewhat larger than this max open book width, as should the support board of the cradle. If the open book occupies a maximum width of 16.13 inches, and the cradle needs to have a maximum travel distance of 4 inches, then the width of the stand must be at least 16.13 + 4 = 20.13 inches.

My stand is 28.75 inches wide. Less the 20 inches for the book gives me about 8 inches to play with. The platen glass and the stand supports eat up this remaining space. This *should* be enough room to fit everything with a little to spare.

Since I can never do trig with pictures to keep track of everything, the second attached image is my triangle.

[megalomania]

In my next step I made a few basic measurements specific to my camera lens. I am using a Canon Rebel XT with a 50mm f/2.5 macro lens. To precisely focus my camera to have a 8.5 x 11 inch piece of paper fill the entire frame, my camera must be about 30.5 inches. This distance is measured from the surface of the paper to the surface plane of the lens (the metal rim).

The camera is focused on the center point of the book or document being photographed. The camera never moves once set, only the book moves, and the center point always remains the same. This is the important principle of building this stand to begin with. Since the max width of books I want to photograph is 8.5 inches, the stand needs to be wide enough to allow the camera to be 30.5 inches away from the book.

As this is a very complicated concept for me to clearly to explain in under a few thousand words, I modeled everything in SolidWorks. (Attached image 1).

The camera and book form a triangle, and the 30.5 inch distance is actually the hypotenuse of this triangle. SolidWorks calculated the actual distance the lens needs to be from the spine of the book as 20.63 inches (this would be the adjacent side of the triangle, which is the width of the base). If the open book is 16 inches wide, that means the spine is only half way, or 8 inches, so the base must be an additional 8 inches wide. This adds up to almost 29 inches.

Does this mean my base now needs to be at least 29 inches wide, plus space for the cradle and platen, plus the thickness of the support arms? Well, no, which I will cover later.

The other lines are for different sized books. I added a book that is 5 inches wide and there is also a 4 inch wide book that I did not label. The dashed lines are what the camera “sees” meaning anything within the triangle of the dashed lines will be photographed by the camera.

Notice that a 5 inch wide book only requires the camera to be 21 inches away from lens to page surface, and the horizontal distance is 14.48 inches. The smaller the book, the closer the camera needs to be moved in to maximize the quality of the image. The larger the book, the farther away the camera needs to be moved.

Why are these measurements important? Because now I know what dimensions to design the camera support arm. My camera needs to slide up and down, and forward and backwards to photograph books of different sizes. I need to design the part that holds the camera to slide forwards and backwards by more than 10 inches. The camera only needs to move up and down a little more than 6 inches.

Example: I just finished photographing a book that was 5 inches wide, a small pocket book. I now want to photograph a large textbook that is 8.5 inches wide. In order to refocus my camera on this bigger book, and to center the margins so I photograph the entire page, I will have to raise my camera by 6.11 inches, and pull it back 9.5 inches.

[megalomania]

I will now explain why the width of the base does not need to be 29 inches or more wide. It may be easier to refer to my earlier design of the camera arm. Notice the camera fits on a tripod head that slides back and forth on a dowel rod. This dowel rod sticks out some 20 inches beyond the max width of the stand. The camera can still be pulled back far enough to be 30.5 inches away from the surface of the book page even though the stand is not that wide.

The threaded rods that attach to the dowel allow the entire works to be raised and lowered. This is the appx 6 inch height change the camera needs to be adjusted by.

I still need to redo the numbers, but I believe the old design allowed the camera to move front and back up to 18 inches, and up and down by up to 8 inches. This should allow a very small document, like a single column of a newspaper, and a quite large document to all fit. Indeed my design accommodates documents of many different sizes, but I only need to worry about designing it to fit 5-8 inch wide books from 0-4 inches thick.

I need all these calculations to figure out where to place the book. If the stand is 29 inches wide, and I have the platen dead center, meaning the spine of the book will be dead center, then the book is only 14.5 inches from the left end of the stand. Referring to the measuring image in my previous post, if the book is 5 inches wide, then the camera (horizontally) is 14.5 away. This means the smallest book I can photograph is 5 inches, more or less.

I say more or less because the lens sticks out about 2-3 inches from the body of the camera. The lens sticks out farther the closer it has to focus, but even pulled all the way back it still juts out as all camera lenses do. Depending on where the body of the camera is attached to the stand, I am limited to books not much smaller than 5 inches.

It would seem 29 inches is just about the minimum width, and I may want a little smaller. At this point it is still too close to tell. This is exactly why I am modeling all this stuff in SolidWorks. The software can calculate the exact dimensions I need so that I do not run into problem during the build. With the design anyway, materials and builder skill are different problems altogether.

[Cobalt.45]

Will adjusting the camera in or out to suit the size of the book, change the apparent size of the font?

[megalomania]

I designed this new camera arm (attached image 1) to fit with the metal stand (attached image 2). The arm bracket uses a short length of square tubing with some L brackets welded on. The square tubing acts like a handle to let me lift the thing. The L bracket to the side has a wooden dowel rod screwed directly onto it at the end. The black wheels are tightening screws to let me raise or lower the camera arm as need be, and then tighten it up so it stays put.

The stand adds a couple of welded bits to either side to let me attach the vertical arms that support the stand and platen. This is as far as I am so far.

Here is my question, and where I may need some assistance:

How can I attach the brackets to the base so they are perfectly perpendicular to the base? I never have welded anything before, and I am about to learn one way or another. I have access to a new electric welder, and an older oxy-acetylene torch. I gather the electric is the one to use for this. My concern is I need those vertical arms shown in the picture to be very straight. The glass platen attaches to these arms and it slides up and down like a train car on railroad tracks. If it is not straight, the thing will not go up and down.

How do I make the metal bracket to attach the vertical bits? The steel tubing is 1 inch square, outside dimensions, so I need the 1 inch tubing to fit inside something that is 1 inch square inside dimensions. Should I weld a couple of L brackets together? Is there some other way to fabricate such a bracket? What would you do?

Should I weld the vertical arms directly to the base, or would having them be detachable be better? Could I just weld some L brackets, or T brackets, to the bottom of the vertical arms and directly bolt it to the base? Would it be stable enough bolting it on, or would it be wobbly?

I may have other problems with my design. The drawer slides that will allow the platen to glide up and down will be attached to the tubing. The slides are slightly wider than one inch, so I plan to add a short standoff so they stick out a little bit from the tubing. The camera bracket has a gap in the front where the L brackets come together so it can wrap around the tubing without hitting the bolts that will hold the drawer slide on.

The problem is, since the bracket will be sliding against the outside of the tubing, how can I attach the drawer slides without having to bolt them from the tubing? It would be easiest to drill two holes straight through the tubing and run the bolt through and tighten it on with a nut. I won’t be able to slide the camera up and down then if the nut is in the way. It would be damn hard to get inside the tubing. I don’t know what to do about that.

I still don’t know how high to make the thing, or in what way to attach the lights. I still need to figure out how far away the lights need to be for even lighting over the surface of the page. They might need to be 12 inches away, or 6 feet high. The arms in my picture are 48 inches long, which I choose at random just to have some starting point.

I am almost certain I am going to have to scrap this whole thing when I reach the end because of some unforeseen design requirement. There is a lot of design stuff dealing with the lights that I will not be able to find out until they lights are on and the book is in place. That is a bad time to find out there is a problem because that means the thing will already be built.

I don’t want to overdesign this thing by making it modular, portable, easy to disassemble, etc because it will never get built. I can’t rightly build it if I don’t have an idea of what skills I have in fabrication, and I don’t have enough money to waste on a mistaken design… Arrgh!

[megalomania]

Will adjusting the camera in or out to suit the size of the book, change the apparent size of the font?

As a mater of fact it will. I find it best to move the camera as close as possible to the book to fill the viewfinder as much as possible with the page because this increases the "DPI" of the scan (I don't have time to go in to why these are not the right terms, for that you will have to read my other posts or my forthcoming book).

The closer the camera is, the better the scan, and the better the OCR you will get. Also, the bigger the page will be if uncorrected.

There is a correction factor you can use with OCR software, or with Photoshop. The DPI of the image file is always the same, but this can be adjusted to correspond with the "life size" dimensions of what you are photographing. I do this by placing a ruler on the inside cover of my books and taking a test shot that gets saved with the rest of the pages.

The size of the book never changes, of course, but the relative size of the book in the photograph changes with the distance of the camera. The actual number of pixels in the image FILE never changes either. By reading the marks on the ruler you can adjust the image size to correspond to whatever pixel dimensions you want.

The attached pictures demonstrate this. In each test shot I moved the camera closer by exactly one inch. The text appears to get larger and larger as the camera zooms in. The area of the page getting photographed decreases, but the number of pixels remains the same. From these I calculated the "DPI" of the pages in each of the shots.

You can see from the pictures the the book is 9 inches wide, and the camera has a horizontal resolution of 3600 pixels. The 401 DPI image is as close as my camera can get without clipping any text, and it is representative of the DPI I would actually get scanning this book. The actual image file is huge because the computer thinks everything is either 300 DPI or 75 DPI. I can then tell Photoshop to resize this image to 401 DPI if I want the picture to be life sized, or any other size I may want. When you put this picture in PDF format, the page should be life sized at 100% zoom if corrected rather than 3 feet wide if it assumes the image is 75 DPI.

All "web" graphics default to 75 DPI because that is the industry standard. This is what I mean when I say using the term DPI to describe these images is not entirely accurate. Most people know what DPI means, so I use it for lack of a better term. The image file, meaning the JPG file format, is, was, and always will be 3600 pixels wide at 75 DPI regardless of the fact I am photographing the moon or a spider. Most people are not concerned with "scale" in their photographs. For text it is only important if you want to very accuratly reproduce a document size.

You can resize a document by guessing. I usually prefer the text to be a little larger, but that is personal choice. If you have, say, a 50 volume set of encyclopedias, and you moved the camera between each book, and you want the final pages to all to be the same size in the final document, you would use this kind of correction.

I have used it in the past because I have never had the benefit of a cradle. Without a cradle, each of my book pages is slightly farther away, meaning the DPI gets less and less with each turn of the page. I think the resolution drops about 15 DPI per inch thereabouts. If I then use photoshop macros to crop all the pages to the same size, and assemble everything into a PDF ebook, if you flip through real fast you can see the pages appear to shrink.

If I instead crop each image based on the margins of the text, the actual dimensions of each successive page will be smaller. In this case resizing all the image files to be the same size will make the text appear to get larger and larger. In fact you are cropping more and more of a 3600 pixel image away and resizing it back to whatever number of pixels. The image may appear to be larger, but the DPI is still going down.

For example, look at the attached images. The widths of the image files are all identical, but the text appears to get smaller. If I cropped the low DPI images to have the same dimensions of the 401 DPI image and resized it them to all be identical size, the 254 DPI image would be "blown up" more.

This problem is precisely one of the reasons I wanted to build this book stand. By keeping the book the same distance from the lens with every photo, the DPI of the book pages remains constant for every page.

[megalomania]

Here are a few shots of me trying to build the stand and failing... I swear the hole in the last image was off by 45 degrees it seemed. The crappy little drill press I had only has about 5 inches of space to fit the work piece in. Between my drill bit (the spade bit in picture 2) and the drill press stand, there is only like 3 inches of room. Not enough space to fit a 3.5 inch board. Doh! The point is moot anyway since the chuck key for the drill press is missing...

[megalomania]

I am nearly done with redesigning my MegaBook scanner. It seems a slight flaw in an angle used to compute the measurements and distances I reported several posts completely ruined most of those estimates. SolidWorks kept the angle of the book at 90 degrees, which screwed up every subsequent calculation (perhaps I didn’t change the angle, we will never know…).

I redid a new calculation part, and I decided to let SolidWorks handle all of the measurements and calculations for angles, distances, etc. The software does a very good job of keeping track of everything and is very precise. Yet another reason I like this software. It seemed more efficient to begin modeling everything and edit it as necessary rather than trying to figure out a bunch of theoretical maximums and minimums.

I made many tweaks and refinements along the way, and I am rather happy with the results so far. There are a few caveats though. I double checked my measurements of lens-to-paper distances, and I added a new measurement of a 10 inch wide piece of paper. The max camera height for this was exactly 34 inches. I modeled three transparent measuring parts using my measurements so I could see where the camera would end up. I also modeled a crude looking, but still dimensionally accurate, model of my camera to see how everything would fit.

I discovered in order to locate the camera 21 inches away from the document (the distance required to focus on a 5 inch wide book), the camera would actually have to be inside the stand. I increased the length of the camera bracket that holds the dowel rod a few inches to allow the camera to clear the stand. I also increased the length of the dowel to 14 inches to allow it to pull back far enough to focus on a 10 inch wide document.

As a consequence of the camera being to the side of the stand side arm, but still needing to focus on the center of the cradle, I moved the side arm forward 2.5 inches. I also moved the cradle itself back an inch. This means the brackets that attach the platen to the stand are not in the center of mass, but are moved forward. This should make lifting the platen a little easier because the handle is in the front, although I may have to weight the handle portion to balance the platen better. Both of these design features are exactly what I wanted to include. The weight of the platen as is may not be sufficient to flatten a book, so adding weight to the thing is a good thing.

I have attached a picture gallery of the assembly as I have designed it right now. A few components are still missing, or not positioned correctly.

The platen, obviously, is not connected to the drawer slides because I am still unsure where and how I want to affix it. This will depend on what materials I have available when the time comes, but will probably be similar to what I have, except the brackets will be wide enough to connect to the slides.

The right support arm of the stand is still in the center, only the left arm is moved forward. I didn’t want to change the right side until I was certain where to put the left side.

The cradle as pictured is not exactly in the center, I shifted it slightly to the left to represent where a single piece of paper would go. The cradle assembly in SolidWorks is moveable, but when I bring anything in as a subassembly it can’t move. It is far easier to just nudge the cradle over rather than edit the original parts. For an actual piece of paper the two V shaped book supports would be together, not with a one inch gap like in the picture. That gap is where the spine of a book would go, a one inch thick book in the case of the picture.

I still have not modeled the lights, nor how the lights will be connected to the rest of the stand. The highest the camera bracket needs to be on the stand is just under 22 inches, even though I currently have the side supports 36 inches high. At some point the lights will attach to the top of those side supports.

I modeled the drawer slides attached to the side support with ¼ inch nuts acting as standoffs. This may actually be unnecessary because the camera bracket at its lowest (for a 5 inch wide book) is still a little shy from hitting the slides. However, I don’t think the model of the slides I used is exactly the same as the slides I have as I grabbed that particular part from a website. It still may be necessary to go a little lower to photograph a document even smaller than 5 inches wide, so it is still a useful to include. In fact, I may increase the length of the standoffs even more to avoid having the brackets on the platen be too big. All the force is put on the point where I weld the bracket to the platen, and the longer those brackets are, the more this (lever) force will be. I wouldn’t want the bracket to snap off after repeated use lifting and lowering the platen.

I still have a few tweaks and measurements to make to make sure the design is valid. I may need to make the glass platen a little smaller since I shrunk the cradle a few inches. I need to make sure the camera bracket can actually hold the weight of the camera sticking out 3 inches like that, and hanging on the end of a 14 inch dowel rod. I also need to figure out how far forward the brackets holding the platen can be moved before it causes problems.

[sbovisjb1]

Do you have anything in the prototype stage? Also the error you may have been experiencing is due to you using a pirated version of Solidworks. High end software like that often put in calculation errors to throw off users who have pirated the software.

[megalomania]

Oh I doubt that, and who says I have a pirated version?

[sbovisjb1]

Its common enough in Software Applications; or so I'm told. I have never seen it myself though.

[ssyd99]

I am also interested in making something similar on lines on Atiz's bookscanner! You post is of extreme value. I will post my design as well soon. In fact I am writing an automated program for controlling the Sony camera which I purchased second hand.

Moderator Note: Obviously Mega's post is of extreme value.... what a stupid thing to say. We would all look forward to seeing your design/program to control the camera.

[megalomania]

I believe some software companies might cripple their software, although they would be wise to improve their security rather than waste their time this way. I meant I don't believe my problem is this. I half jokingly meant there was a slight "flaw," that being I I started it as a quick mockup measuring part and forgot to change the angle because... It snapped the line to 90 degrees automatically, which is usually what you want, just not this time.

[hereticalhermit]

I have tried some of the methods mentioned above, and have found the following:
(1) The best resolution to make this work seems to be anything at or above 3.1 megapixels. The next lower setting on my camera is 1.8 mp, at which level I cannot read the text on my computer screen in order to apply any corrections. It may also be that my computer is old and inexpensive, and that a newer one with a better screen might correct this.
(2) A large and heavy glass holds down the book well enough, but the iron in the glass which gives it a greenish tint also makes for darker photos.
(3) The lights mounted at angles to the glass and above it will prevent glare. The bulbs do not need to be anything bigger than 2 or 3 60w bulbs.
(4) The glass needs to be lifted in order to turn each page. This necessitates using latex gloves to prevent finger prints and smudges.
I have not had any success with ocr software, but have only tried the cheap (that is, free) stuff available over the internet. Better stuff is available for a fee. If I knew more about computers and software, I might have better success.
Hopefully, this will be helpful to some of us out there.

[megalomania]

I have to concur on point 1. I have a post about minimum resolution somewhere, and I have a few paragraphs in my earlier guide (which I am no longer circulating while I revise it). I tested this about 5 years ago when high resolution cameras were hard to come by at a reasonable price. 3MP is the bare minimum for readable text. To go lower is possible only by photographing a small area, but some cheap cameras with cheap lenses can't focus up close.

Point 2 explains why my earlier pictures looked greenish, but I exclusively decolorize everything I do. There is a color special card you can use to assist in color correction of the images. This technique is used for accurate color reproduction for paintings, or colored images, by professionals. I use my camera's custom color correction to reproduce white as white, but this is to correct color differences caused by your light source.

As for point 3, you certainly need to angle the lights, but the intensity of light varies with camera models, and with camera settings. If you increase your exposure times you can let more light into the camera, the same with widening the aperature. I find even a room light is adequate for reading the text in the original image, but because I decolorize and convert images to bitonal, it is very hard to get results unless a lot of light is used. A strong light source makes a better contrast between the black of text and the whiteish/yellowish of the background, and this makes it easier to preserve only the text.

I make images into bitonal because it gets the smallest possible filesize in the resulting pdf document. I learned the hard way that a little shadow, or a weak light source has a big impact on the post edited images. Sometimes you can't see anything wrong with the original color photo, but when you end up with an entire book ruined because a camera strap cast a slight shadow because you accidentially left the ceiling light on, you start to consider light sources as important. That shadow resulted in a solid black mass in the middle of all my pages. It defeats the purpose of using a camera if you have to spend so much time manually correct the images.

As for point 4, I use an oversized piece of glass that I only touch at the edges to lift. I also keep a chamios cloth handy to wipe the dust and fingerprints off. Cloth inspectors gloves would actually be better for glass. I have been meaning to try using one of those rubber fingers to turn the pages easier.

I didn't know there was free OCR software. I am not suprised there is, but nothing compares with Finereader or Omnipage. Both of these are almost identical, performance wise, but vary only by user interface. Basically it is personal preference. Both are "free" over at mininova.org.

[sbovisjb1]

And all those pesky "Stop pirating" messages you Americas keep getting for nicking software will go away soon, as a lot of software patents are about to be revoked.

[FUTI]

OCR software makes lower number of errors only at and above of 200dpi. Crude conversion to A4 size paper give you a about 2MP. Just to be sure and to put things to the safe side I wouldn't use nothing less then around 3MP.

[megalomania]

Check out my pics in post 70. What dpi you get depends on the lens distance to the page and # of megapixels. Indeed anything less than 200 dpi is garbage when ocr'd.

I am working on a technique using panorama software to combine multiple shots of pages into one in order to increase resolution. You can nearly double you resolution by taking a pic of half the page and using the software to combine it into a single picture.

The technique works fine with manual correction, but I am looking for the best automated solution. This technique is more applicable to works that need to be preserved at high resolution and oversized pages, not for regular book scanning. I would use this on books I want to preserve, not just copy, because of the time involved.

However, this type of software gets better and better over time. I may just find one that actually does a good job. The useful part of this is you could use two cheaper camera taking pictures at the same time and merge the images. You could get a 10-14 MP image from a pair of cheapo 8 MP cameras. If you really wanted super resolution you could do 4x cameras at once.

I first tried this several years when I book I was scanning had a foldout map. The software did a good job, but it left a noticeable seam where the overlapping sentences merged together that had to be edited out in photoshop.

It takes double the time to photograph a book using this method with only a single camera, and even more time to post process the images unless automation is used. Since all cameras with fixed lenses hit a ceiling at 10MP, the only chance to get higher resolution is to use two or more cameras at a time. Before someone asks, I know there are some digital cameras above 10 megapixels, but the cheapo lenses of fixed lens cameras can only reproduce images of up to 10 megapixels in quality. Consumer point-and-shoot cameras with more than 10 MP are just a marketing ploy to sell more cameras. Only a dSLR camera with a quality lens can utilize high resolution sensors above 10 MP, but they cost a fortune. I have said this before, the lens on my dSLR camera costs more than most regular digital cameras; it is the lens, not the camera, that gets the best image of book pages.